Evolutions in AI: what to expect in 2022?

AI or artificial intelligence is a leading sector in the IT industry. With a growth rate of 15.2% in 2021, artificial intelligence, deep learning and machine learning solutions and platforms have profoundly changed the organisation of work and the relationship between man and machine. But AI is also impacting many other areas: sales, marketing, science, industry, etc. According to Sundar Pichai, CEO of Google Inc, the impact of artificial intelligence will be even greater than that of fire and electricity on humanity. This may sound like a bit of an exaggeration, but AI technologies are indeed disrupting technological, economic and societal environments.

And with a new growth rate of 18.8% forecast for 2022, artificial intelligence will continue to evolve and this will lead to more changes.

Find out in this article the trends in AI evolution for 2022.

An expansion of AI models 

Deep learning, artificial intelligence and machine learning models rely on the collection and analysis of large amounts of data. In 2022, these models are expected to continue to expand and rely on even larger data sets to make increasingly accurate decisions.

This expansion is largely due to the continued evolution of GPT (Generative Pre-Training) which creates unsupervised language models based on Transformer models. 

These transformers are themselves deep learning models used mainly as a technique in NLP (Natural Language Processing), the technology that allows machines to “understand” human language.

In this field, the latest model, currently in beta version and expected to be available in 2022, is GPT-3. It is 100 times larger than its predecessor (GPT-2). With 175 billion parameters, the GPT-3 achieves high scores in various NLP benchmarks. In particular, it is said to be able to write and edit articles that are considered indistinguishable from those written by humans. 

Stronger links between AI and cybercrime

Cybercrime has increased exponentially in 2021, with over 330,000 additional cyberattacks. These numbers are likely to increase further in 2022. AI plays a major role in this increase as connected objects and smart machines multiply in our daily lives, increasing the potential attack surfaces, vulnerabilities and loopholes. The rationale is clear: as the number of devices added to a network increases, so does the number of vulnerabilities that cyber attackers can exploit to access data.

Artificial intelligence can also be exploited by cyber attackers to design more sophisticated attacks such as:

• automated phishing which relies on automation and deep learning processes to increase the effectiveness of phishing by personalising messages;

• hacking AI-driven devices such as autonomous cars, military drones, etc.

• the writing of fake articles and rumours by AI algorithms. AI can also be exploited to construct false evidence;

• corruption of large volumes of data by “poisoning” AI models;

• learning-based cyberattacks, which include the massive use of AI to probe for weaknesses in systems before launching multiple simultaneous attacks.

However, the developments in AI for 2022 show that it will also play an important role in preventing these cyber threats. In particular, AI can identify patterns, analyse network traffic and highlight suspicious activity through intelligent algorithms. AI will also continue to help prioritise alerts and quickly visualise the urgency of threats. 

Finally, in security as in many other areas, technological innovation is driven by artificial intelligence. Passwords will increasingly be replaced by biometrics and authentication techniques based on behavioural data. Machine learning will become a proactive method of creating ever more robust cyber security protocols.

A transformation of the labour market with AI developments

AI is often associated with the fear that machines or robots will eventually replace the human workforce. However, the figures for 2021 show that the opposite is true. Companies and organisations are using AI technologies to analyse, interpret and extract information from data. The use of these technologies significantly increases their need for human labour to implement, maintain and secure them.

AI has thus taken over many departments of the company, from engineering and production lines to security and marketing. But for each sector, it has had to integrate with human organisations and teams to facilitate their work and increase their productivity.

According to “The Future of Jobs Report” published in 2018, AI is also expected to create more than 58 million jobs by 2022. However, not all jobs would be equal in the face of this development. Some jobs that can be easily automated may disappear, such as certain secretarial tasks, information management tools, data entry operators, etc. However, others will experience a growing demand, such as data scientists, computer scientists specialising in artificial intelligence, etc.

To accompany this transformation of the labour market linked to developments in AI, France has signed an agreement for 2022 providing for the creation of a laboratory dedicated to artificial intelligence: LaborIA. This initiative is part of the second GPAI (Global Partnership on AI) summit, which supports cutting-edge research and implementation activities on artificial intelligence priorities.

LaborIA’s objectives are to identify the challenges and use of AI by businesses and administrations. Its role is to deploy concrete experiments in the workplace to detect difficulties and create debates on the use of AI technologies in a professional environment.

Sources and useful links:

International Data Corporation (IDC) report: https://www.idc.com/tracker/showproductinfo.jsp?containerId=IDC_P37251

OpenAI’s GPT-3: https://openai.com/blog/openai-api/

Presentation and objectives of LaborIA: https://travail-emploi.gouv.fr/actualites/l-actualite-du-ministere/article/laboria-centre-de-ressources-et-d-experimentations-sur-l-intelligence